figure b

Introduction

Whole-grain foods are widely recommended as a preferred source of carbohydrate for people with diabetes [1, 2]. Regular consumption of whole grains has been associated with a reduced risk of developing type 2 diabetes, and improved blood glucose or other cardiovascular risk factors in those with diabetes [3, 4]. Recent reports, however, have indicated that food processing techniques, such as milling, attenuate the blood glucose improvements expected from whole-grain consumption [5, 6].

Although starch is the largest contributor to energy within the diet [7], starch structure and the nature of starch is not yet considered in advice for those with diabetes [1]. Native starches are present within whole grains, but many lose structure and gelatinise when heated with water. There is evidence in adults with normal glucose tolerance that gelatinised starch is digested more rapidly than native starch [8].

In several countries, breakfast cereals are a primary source of whole grains [9], and may vary by extent of milling or whether the grains have been cooked to achieve starch gelatinisation. In this study, we examined how milling and cooking of whole grains influence their effect on blood glucose using four whole-wheat breakfast meals which differed by preparation method and grain milling.

Methods

This randomised crossover study was approved by the New Zealand Health and Disability Ethics Committee (17/STH/41) and prospectively registered (ACTRN12617000328370). Informed consent was obtained from all participants.

Individuals with type 2 diabetes aged 18–75 years were eligible to participate if their weight was stable and their medication had not changed over the past three months. Exclusion criteria included pregnancy, lactation, wheat or gluten allergies and coeliac disease. Participants were recruited through advertising in community groups and diabetes clinics.

On separate days, participants received four test-meals and three glucose standards in a randomised order (computer-generated). Each sequence was sealed in an opaque envelope. Established postprandial blood glucose testing protocols described in detail elsewhere [5] were followed regarding participant preparation, testing frequency and test-meal administration [10].

The test-meals consumed were a wet whole-wheat porridge balanced for ingredients and not commercially available. Four test-meals were administered, consumed hot or cold, with either finely milled (particle size <150 μm) or coarse kibbled whole-wheat (particle size ≥1680 μm) (Champion Flour, Christchurch, New Zealand). All meals were prepared with natural unsweetened yoghurt and water then sealed in foil packages for at least 12 h. Each test-meal provided 1300 kJ of energy. On test days, meals were either cooked to >85°C for 15 min to gelatinise starch and served at ≥65°C to prevent retrogradation, or uncooked and served cold (10–15°C). Each test-meal and glucose standard provided 50 g of available carbohydrate.

Statistical analysis

Our sample size estimate of 17 adults was calculated with a within-person correlation of 0.3 and power of 0.90 to detect a one SD difference in postprandial blood glucose between test-meals, as measured by the incremental AUC (iAUC). iAUC was calculated over 3 h with the trapezoidal method ignoring the area below the baseline [11]. Data were analysed with mixed-effects regression models using a robust variance structure to account for heteroskedasticity. The predictor variable was binary for starch gelatinisation (yes/no) or particle size (fine/kibbled). Participant ID was included as a random effect to account for repeated measures. Analyses were performed in Stata 15.1 (StataCorp, TX, USA), and data are expressed as mean differences (MD) with 95% CI. The statistician was blinded to intervention.

Results

Eighteen participants [mean age 63.1 ± 9.8 (range 36–75) years, HbA1c 57.0 ± 11.5 mmol/mol (7.4 ± 3.2%), BMI 33.0 ± 7.5 kg/m2, diabetes duration 9.5 years] were randomised and completed the study between June 2017 and January 2018. Seven (39%) participants were female. Four (22%) participants identified as Māori, three (17%) as Pacific peoples, one (6%) as Sri Lankan and ten (55%) as New Zealand European. Four participants dropped out of the study.

Results from test-meals and glucose standards are shown in Table 1. Postprandial blood glucose was 59% lower (95% CI 42, 77%) after consuming uncooked, ungelatinised (native starch) whole-wheat porridge compared with cooked, gelatinised whole-wheat porridge. Similarly, postprandial blood glucose was 43% lower (95% CI 20, 66%) after consuming more intact whole-wheat than consuming finely milled whole-wheat. The overall iAUC difference was 268 mmol/l × min (95% CI 188, 348 mmol/l × min) between gelatinised and native starch, and 173 mmol/l × min (95% CI 80, 266 mmol/l × min) between finely milled and coarse kibbled whole-wheat. The p value for an interaction between nature of starch and whole-wheat particle size was 0.841. No adverse events were reported.

Table 1 Differences in postprandial blood glucose due to starch gelatinisation and whole-wheat milling in whole-grain porridge (n=18)
Full size table

Discussion

This study has shown that two commonly applied food processing techniques, starch gelatinisation and milling, have the potential to independently influence blood glucose levels following the consumption of whole grains by adults with type 2 diabetes.

Numerous studies in normoglycaemic individuals have reported that starch gelatinisation is associated with a greater glycaemic response than occurs when comparable amounts of native starch are consumed in raw foods or cooked foods eaten cold or at room temperature [12,13,14]. However, we believe this may be the first demonstration of this effect in people with diabetes, who experienced a more marked glycaemic response than that observed in normoglycaemic individuals.

The findings of this study are complementary to recent work showing that fine milling of whole grains is associated with greater postprandial blood glucose than when more intact whole grains are consumed [5, 6]. Starch gelatinisation and milling are applied widely to carbohydrate-containing foods and contribute to the more rapid and complete digestion of starch. This is likely due to increased exposure of starch to the action of salivary and pancreatic amylases [15].

This study is limited by the fact that it describes only the immediate consequences of processing whole grains. However, the striking effects of heating and milling on postprandial blood glucose, an accepted determinant of overall glycaemic control and cardiovascular risk [16], suggest that these results warrant incorporation into nutrition recommendations and in dietary counselling. Current nutrition recommendations support the use of whole grains as an appropriate source of dietary carbohydrate for people with diabetes. This study indicates that breakfast cereals will have a reduced glycaemic response if they contain largely intact whole grains, and if consumed as a ‘muesli type’ product or soaked overnight rather than as hot, cooked porridge. The extent to which milling of whole-grain wheat influences glycaemic response suggests that the recommendations for patients with diabetes to choose whole grains over refined grains need to be reconsidered. Definitions for whole-grain foods [17, 18], especially for foods with health claims related to diabetes and glucose control, should factor in the impact of grain milling.